Keywords: Visualization, 4D-6D Principal Component Domains, Conditioning and Binning, Protein Folding Simulation
Limitations of human perception and cognition restrict the number of domain variables that can be usefully represented in function visualization. Common practice limits the function domain to two variables. By makes resolution compromises, production of 4-D domain visualizations is straight forward and can be informative. The approach here uses two-way conditioned panel plots provide a low a resolution encoding of two variables. Higher-resolution hexagon bin cell centers in each panel encode two more variables. Color or ray glyphs in a hexagon bin cells encode one or a few dependent variable statistics for each cell. In the simulated protein folding examples, the domain variables are principal components are computed from carbon-alpha atomic coordinates and the dependent variable is energy. In publications, color has worked well for drawing attention to low energy wells as they change from panel to panel and for showing point locations of experimentally known structures. Higher dimensional extensions use 3-way conditioning, 3D binning or both. The conditioning is simple. The soon to be released R package, TObin, provides truncated octahedron binning. Glisten, 3D rendering software, provides separate widgets for interacting with each 3D graphics layer. This helps to address occlusion problems.